Research map / Vision-language and multimodal / Multimodal LLMs
Token-Efficient VLM: High-Resolution Image Understanding via Dynamic Region Proposal
Yitong Jiang, Jinwei Gu, Tianfan Xue, Ka Chun Cheung, Pavlo Molchanov, Hongxu Yin, Sifei Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add 473 dynamic patches to 256 static patches (ablation) | TEVA-3B · V* | accuracy: 45.0 -> 61.6 | Improved |
| add 473 dynamic patches to 256 static patches (ablation) | TEVA-3B · TextVQA | accuracy: 44.5 -> 66.8 | Improved |
| add 473 dynamic patches to 256 static patches (ablation) | TEVA-3B · POPE | accuracy: 83.0 -> 87.0 | Improved |
| add 473 dynamic patches to 256 static patches (ablation) | TEVA-3B · MME-RealWorld | accuracy: 30.0 -> 34.0 | Improved |
| add 80 dynamic patches to 81 static patches (ablation) | TEVA-3B · TextVQA | accuracy: 44.5 -> 54.2 | Improved |
| add Subject-Oriented pretraining data (ablation) | TEVA-3B · MME-RealWorld reasoning | accuracy: 28.5 -> 34.3 | Improved |
| add Subject-Oriented pretraining data (ablation) | TEVA-3B · TextVQA | accuracy: 58.7 -> 66.8 | Improved |
| add Subject-Oriented pretraining data (ablation) | TEVA-3B · MME-RealWorld perception | accuracy: 33.1 -> 33.8 | No clear effect |
| add Subject-Oriented pretraining data (ablation) | TEVA-3B · POPE | accuracy: 86.5 -> 87.0 | No clear effect |
| add affine transformation matrix embedding (ablation) | TEVA-3B · V* SRR | accuracy: 50.0 -> 63.2 | Improved |
| add affine transformation matrix embedding (ablation) | TEVA-3B · V* AR | accuracy: 48.3 -> 60.0 | Improved |
| use dynamic patch sampling instead of multiple vision tower passes (vs V*) | TEVA-7B · V* full pipeline inference | relative time: 12.31x -> 1.00x | Improved |
Closest papers by meaning
- Variation-aware Vision Token Dropping for Faster Large Vision-Language Models (CVPR 2026 · similarity 0.90)
- A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs (CVPR 2025 · similarity 0.90)
- Inference Optimal VLMs Need Fewer Visual Tokens and More Parameters (ICLR 2025 · similarity 0.90)
- iLLaVA: An Image is Worth Fewer Than 1/3 Input Tokens in Large Multimodal Models (ICLR 2026 · similarity 0.90)
- SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference (ICML 2025 · similarity 0.90)
- ApET: Approximation-Error Guided Token Compression for Efficient VLMs (CVPR 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
